OPTION(mds_beacon_grace, OPT_FLOAT, 15)
OPTION(mds_enforce_unique_name, OPT_BOOL, true)
OPTION(mds_blacklist_interval, OPT_FLOAT, 24.0*60.0) // how long to blacklist failed nodes
+
OPTION(mds_session_timeout, OPT_FLOAT, 60) // cap bits and leases time out if client idle
+OPTION(mds_session_blacklist_on_timeout, OPT_BOOL, true) // whether to blacklist clients whose sessions are dropped due to timeout
+OPTION(mds_session_blacklist_on_evict, OPT_BOOL, true) // whether to blacklist clients whose sessions are dropped via admin commands
+
OPTION(mds_sessionmap_keys_per_op, OPT_U32, 1024) // how many sessions should I try to load/store in a single OMAP operation?
OPTION(mds_revoke_cap_timeout, OPT_FLOAT, 60) // detect clients which aren't revoking caps
OPTION(mds_recall_state_timeout, OPT_FLOAT, 60) // detect clients which aren't trimming caps
}
mds_lock.Lock();
- stringstream dss;
- bool killed = kill_session(strtol(client_id.c_str(), 0, 10), true, dss);
+ std::stringstream dss;
+ bool killed = kill_session(strtol(client_id.c_str(), 0, 10), true,
+ g_conf->mds_session_blacklist_on_evict, dss);
if (!killed) {
dout(15) << dss.str() << dendl;
ss << dss.str();
C_GatherBuilder gather(g_ceph_context, reply);
for (const auto s : victims) {
- server->kill_session(s, gather.new_sub());
+ std::stringstream ss;
+ kill_session(s->info.inst.name.num(), false,
+ g_conf->mds_session_blacklist_on_evict, ss, gather.new_sub());
}
gather.activate();
}
objecter->maybe_request_map();
}
-bool MDSRankDispatcher::kill_session(int64_t session_id, bool wait, std::stringstream& err_ss)
+bool MDSRank::kill_session(int64_t session_id,
+ bool wait, bool blacklist, std::stringstream& err_ss,
+ Context *on_killed)
{
+ assert(mds_lock.is_locked_by_me());
+
+ // Mutually exclusive args
+ assert(!(wait && on_killed != nullptr));
+
if (is_any_replay()) {
err_ss << "MDS is replaying log";
return false;
std::string tmp = ss.str();
std::vector<std::string> cmd = {tmp};
- C_SaferCond on_blacklist_inline;
- Context *on_blacklist_done = nullptr;
- if (wait) {
- on_blacklist_done = &on_blacklist_inline;
- } else {
- on_blacklist_done = new FunctionContext([this, session_id](int r) {
- objecter->wait_for_latest_osdmap(
- new FunctionContext([this, session_id](int r) {
- auto epoch = objecter->with_osdmap([](const OSDMap &o){
- return o.get_epoch();
- });
-
- set_osd_epoch_barrier(epoch);
-
- Session *session = sessionmap.get_session(
- entity_name_t(CEPH_ENTITY_TYPE_CLIENT, session_id));
- if (session) {
- server->kill_session(session, NULL);
- } else {
- dout(1) << "session " << session_id << " was removed while we waited "
- "for blacklist" << dendl;
- }
- }));
- });
- }
+ auto kill_mds_session = [this, session_id, on_killed](){
+ assert(mds_lock.is_locked_by_me());
+ Session *session = sessionmap.get_session(
+ entity_name_t(CEPH_ENTITY_TYPE_CLIENT, session_id));
+ if (session) {
+ if (on_killed) {
+ server->kill_session(session, on_killed);
+ } else {
+ C_SaferCond on_safe;
+ server->kill_session(session, &on_safe);
- dout(4) << "Sending mon blacklist command: " << cmd[0] << dendl;
- monc->start_mon_command(cmd, {}, nullptr, nullptr, on_blacklist_done);
+ mds_lock.Unlock();
+ on_safe.wait();
+ mds_lock.Lock();
+ }
+ } else {
+ dout(1) << "session " << session_id << " was removed while we waited "
+ "for blacklist" << dendl;
- if (wait) {
- mds_lock.Unlock();
- int r = on_blacklist_inline.wait();
- mds_lock.Lock();
- if (r != 0) {
- err_ss << "Failed to blacklist, r=" << r;
- return false;
+ // Even though it wasn't us that removed it, kick our completion
+ // as the session has been removed.
+ if (on_killed) {
+ on_killed->complete(0);
+ }
}
+ };
+
+ auto background_blacklist = [this, session_id, cmd](std::function<void ()> fn){
+ assert(mds_lock.is_locked_by_me());
+
+ Context *on_blacklist_done = new FunctionContext([this, session_id, fn](int r) {
+ objecter->wait_for_latest_osdmap(
+ new C_OnFinisher(
+ new FunctionContext([this, session_id, fn](int r) {
+ Mutex::Locker l(mds_lock);
+ auto epoch = objecter->with_osdmap([](const OSDMap &o){
+ return o.get_epoch();
+ });
+
+ set_osd_epoch_barrier(epoch);
+
+ fn();
+ }), finisher)
+ );
+ });
+
+ dout(4) << "Sending mon blacklist command: " << cmd[0] << dendl;
+ monc->start_mon_command(cmd, {}, nullptr, nullptr, on_blacklist_done);
+ };
- // Wait for latest OSDMap so that we can learn the OSDMap epoch in
- // which the blacklist happened
- C_SaferCond on_latest_osdmap;
- objecter->wait_for_latest_osdmap(&on_latest_osdmap);
+ auto blocking_blacklist = [this, cmd, &err_ss, background_blacklist](){
+ C_SaferCond inline_ctx;
+ background_blacklist([&inline_ctx](){inline_ctx.complete(0);});
mds_lock.Unlock();
- on_latest_osdmap.wait();
+ inline_ctx.wait();
mds_lock.Lock();
- auto epoch = objecter->with_osdmap([](const OSDMap &o){
- return o.get_epoch();
- });
+ };
- set_osd_epoch_barrier(epoch);
+ if (wait) {
+ if (blacklist) {
+ blocking_blacklist();
+ }
// We dropped mds_lock, so check that session still exists
session = sessionmap.get_session(entity_name_t(CEPH_ENTITY_TYPE_CLIENT,
"for blacklist" << dendl;
return true;
}
-
- C_SaferCond on_safe;
- server->kill_session(session, &on_safe);
-
- mds_lock.Unlock();
- on_safe.wait();
- mds_lock.Lock();
+ kill_mds_session();
+ } else {
+ if (blacklist) {
+ background_blacklist(kill_mds_session);
+ } else {
+ kill_mds_session();
+ }
}
return true;
projects / ceph.git / commitdiff